Identification of Trigeminal Sensory Neuronal Types Innervating Masseter Muscle

Abstract Understanding masseter muscle (MM) innervation is critical for the study of cell-specific mechanisms of pain induced by temporomandibular disorder (TMDs) or after facial surgery. Here, we identified trigeminal (TG) sensory neuronal subtypes (MM TG neurons) innervating MM fibers, masseteric fascia, tendons, and adjusted tissues. A combination of patch clamp electrophysiology and immunohistochemistry (IHC) on TG neurons back-traced from reporter mouse MM found nine distinct subtypes of MM TG neurons. Of these neurons, 24% belonged to non-peptidergic IB-4+/TRPA1– or IB-4+/TRPA1+ groups, while two TRPV1+ small-sized neuronal groups were classified as peptidergic/CGRP+. One small-sized CGRP+ neuronal group had a unique electrophysiological profile and were recorded from Nav1.8– or trkC+ neurons. The remaining CGRP+ neurons were medium-sized, could be divided into Nav1.8–/trkC– and Nav1.8low/trkC+ clusters, and showed large 5HT-induced current. The final two MM TG neuronal groups were trkC+ and had no Nav1.8 and CGRP. Among MM TG neurons, TRPV1+/CGRP– (somatostatin+), tyrosine hydroxylase (TH)+ (C-LTMR), TRPM8+, MrgprA3+, or trkB+ (Aδ-LTMR) subtypes have not been detected. Masseteric muscle fibers, tendons and masseteric fascia in mice and the common marmoset, a new world monkey, were exclusively innervated by either CGRP+/NFH+ or CGRP–/NFH+ medium-to-large neurons, which we found using a Nav1.8-YFP reporter, and labeling with CGRP, TRPV1, neurofilament heavy chain (NFH) and pgp9.5 antibodies. These nerves were mainly distributed in tendon and at junctions of deep-middle-superficial parts of MM. Overall, the data presented here demonstrates that MM is innervated by a distinct subset of TG neurons, which have unique characteristics and innervation patterns.

MM consists of three heads (superficial, middle, and deep) of muscle fibers, masseteric fascia, and tendons; and is innervated by the masseteric nerve, a branch of the anterior portion of the mandibular division (V3) of the trigeminal (TG) nerve (Fig. 13A,B). Anatomical studies have revealed myofascial trigger points, which represent tender muscle areas that can elicit pain whenever stimulated during normal or pathologic condition. These myofascial trigger points coincide with neuromuscular junctions at the innervation zone between MM heads (Procopio Pinheiro et al., 2020). Previous observations of the head and neck area showed the presence of nociceptive unmyelinated C-and myelinated Ad fibers (Strassman et al., 1996;Strassman and Raymond, 1999), as well as low-threshold mechanoreceptors (LTMRs) consisting of myelinated Aa and Ab fibers (Lobbezoo et al., 2002). Similarly, it has been shown that MM has C-fibers, two types of A-d highthreshold mechanoreceptors (Ad -HTMR) as well as Ab -LTMR fibers (Nishimori et al., 1986;Cruccu et al., 1989;Connor et al., 2005;Wong et al., 2014;Sato et al., 2018). MM is predominantly innervated by Ad -HTMR, with cell bodies observed to be 27% small, 49% medium, and 24% large-sized neurons (Sato et al., 2018). However, the precise composition of MM innervating TG ganglion neuronal groups or their properties is unknown.
The understanding of molecular mechanisms governing masticatory myofascial pain is hindered by the limited information on the neuronal circuits undergoing plasticity during TMD and after surgical procedures involving MM and/or masseteric nerve. The aim of the present study has been to identify and comprehensively characterize TG sensory neuronal groups innervating MM (MM TG neurons), and to examine the innervation patterns of MM by afferent sensory fibers. To do so, back-tracing of sensory neurons from MM, patch-clamp electrophysiology, Ca imaging, anatomic studies, sensory neuronal reporter mice and tissues from a nonhuman primate species (common marmosets) have been used (Herrity et al., 2014;Patil et al., 2018).

Animals
All animal experiments conformed to American Pain Society (APS)'s Guiding Principles in the Care and Use of Vertebrate Animals in Research and Training. We also followed guidelines issued by the National Institutes of Health (NIH) and the Society for Neuroscience (SfN) to minimize the numbers of animals used and their suffering. Protocols used in this study (20190114AR for mice and 20200021AR for nonhuman primates) are approved by the Institution Animal Care and Use Committee (IACUC).
Two aged (11-and 15-year-old) male common marmosets (Callithrix jacchus) were used for collection of MM. IACUC and veterinary oversight regularly monitored marmoset housing and animal conditions to ensure all guidelines for the health and safety of the animals were met and research was conducted in compliance with the United States Public Health Service's Policy on Humane Care and Use of Laboratory Animals and the Guide for the Care and Use of Laboratory Animals and adhered to the American Society of Primatologists (ASP) principles for the ethical treatment of non-human primates. Marmosets were euthanized for humane reasons following veterinary consult and tissue samples were collected immediately following determination of death.

Primary TG neuronal culture
To visualize MM TG neurons, wild-type or reporter mice expressing GFP, YFP (Ai32), or tdTomato (Ai9) gene were injected into the right and left MM closer to tendon with 10 ml of WGA-488 or WGA-555 (50 mg in 0.5% DMSO) back-tracer. Neuronal culture for electrophysiology recordings was always generated from one mouse. TG was dissected 24-36 h post-WGA injections, and sensory neurons were cultured as previously described (Belugin et al., 2013). Cells were maintained in DMEM with 2% fetal Figure 1. Recordings from non-peptidergic IB4 1 MM TG neurons. A, WGA-488 1 /IB4-555 1 (marked with blue arrows), but not WGA-488 1 /IB4 -(marked with yellow arrows) were selected for recording non-peptidergic TG neurons innervating MM. B, Stimulus waveform (1 nA, 0.5 ms) indicated below trace generated a single AP in a WGA 1 /IB4 1 TG neuron belonging to the S2 group (Table  1). AP width is duration at base, dB. AHP80 is the time required for the AHP (measured in mV) to decay by 80% to a RMP level. Characteristic AP deflection is indicated by black arrow. C, AP from a WGA 1 /IB4 1 TG neuron belonging to the group S1 (Table 1). Distinctive AP feature, bow, is indicated by black arrow. Distance from RMP to a lowest point of AP, AHP peak, is measured as indicated with the green arrow. D, Current-evoked AP train from a WGA 1 /IB4 1 TG neuron belonging to the S2 group. Current waveforms are below trace and are applied by steps from 50 to 550 pA with 100-pA increment. Depictured AP train is evoked by a 150-pA step lasting 1 s. E, Currents were generated from a WGA 1 /IB4 1 TG neurons belonging to the S1 group by the indicated waveforms found below traces. The decay constant t was derived from standard single exponential fits between points indicated by arrows for the outward portion of the final current trace (120 mV). bovine serum, 2 mM L-glutamine, 100 U/ml penicillin, and 100 mg/ml streptomycin. No growth factor was added to the media. The experiments were performed within 24 h after TG neuron plating. This culture conditions minimize changes in sensory neurons (Patil et al., 2018).
Recordings were made in patch clamp whole-cell voltage [holding potential (V h ) of -60 mV] or current clamp configurations at room temperature. Data were acquired using an Axopatch 200B amplifier and analyzed with pCLAMP10.6 software (Molecular Devices). Recording data were filtered at 0.5-5 kHz and sampled at 2-20 kHz depending on current kinetics. Borosilicate pipettes (Sutter) were polished to resistances of ,5 MV. If required, access resistance (R s ) was compensated (40-80%) to the value of ,6-8 MV. Data were rejected when R s changed .20% during recording, leak currents were .100 pA, or input resistance was ,300 MV. Liquid junction potential (LJP) was not corrected using the V meter = V cell 1 LJP equation, since large anions were not included in extracellular or pipette recording solutions. Recordings were made 3-4 min after establishing wholecell configuration to allow for equilibrium. Currents were considered positive when their amplitudes were fivefold bigger than displayed noise (in root mean square).
Standard external solution (SES) contained the following: 140 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM D-glucose, and 10 mM HEPES, pH 7.4. The standard pipette solution (SIS) contained the following: 140 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM EGTA, 10 mM D-glucose, and 10 mM HEPES, pH 7.3, 2.5 ATP and 0.2 GTP. Drugs were applied by a fast, pressure-driven and computer controlled four-channel system (ValveLink8; AutoMate Scientific) with quartz application pipettes. Recordings were independently performed by two investigators. Data were accumulated from five to eight independent MM TG neuronal cultures for each mouse line. Six to 15 neurons were recorded from each MM-TG neuronal culture.

Electrophysiology: recording protocols and data analysis
On the selected for recording MM TG neurons, we used a sequence of protocols after fast and slow capacitance compensations: (1) single action potential (AP) in currentclamp configuration was generated with 1 nA (2 nA for .40-pF cells) 0.5-ms current pulse (Fig. 1B,C; Petruska et al., 2000;Patil et al., 2018); (2) AP train was induced by applying step currents of 50-550 pA with 100-pA increment for 1 s (Fig. 1D) or 200-2000 pA with 300-pA increment for medium-to-large neurons; (3) after current clamp recordings the electronics were switched to voltageclamp configuration (V h = -60 mV) and ATP (30 mM) current was recorded by applying drug for 5 s; (4) after ATPgated current, 5HT (30 mM)-gated current was recorded by 5-s-long drug application; (5) the next protocol in a voltage-clamp configuration was a step down from V h to -80 mV kept for 500 ms, and then 200-ms depolarizing command steps (20 mV) were applied from -40 mV to a final potential of 120 mV ( Fig. 1E; Petruska et al., 2000). Time gap between these successive one to five protocols was 1-2 min. In some sets of experiments, mustard oil (MO; 25 mM) and menthol (Ment; 100 mM) responses in MM TG neurons of reporter mice were evaluated using Ca 21 imaging system as previously described (Salas et al., 2009;Patil et al., 2018).
Data on sensory neuronal marker expression in MM TG neurons, capacitance (in pF) and resting membrane potential (RMP; V m in mV) values were collected prior recording and after application of protocol-1 (Fig. 1A,B). Cells were considered as a spherical, and therefore, diameter (d in mm) of cells was calculated from capacitance (C m in pF) values using the following: AP duration at the base (dB; time from V m starting point to V m levels at falling phase of AP), after-hyperpolarization (AHP) peak as a distance from RMP to a lowest point of AP and 80% recovery time of AHP to baseline (AHP 80 ) were measured from data generated by protocol-1 (  Table 1). Besides these AP parameters, we noted characteristic features of AP shapes, such as "hump," "bow," "deflection," and "straight," on the falling phase of AP (Patil et al., 2018;Figs. 1B,C, 2A,B, 3B,C; Table 1). Analysis of protocol-2 gave AP activation threshold (in pA) and maximum AP frequency at the given current input ( Fig. 1D; Table 1). Protocol-3 and protocol-4 revealed algesic responses to ATP and 5HT as well as I ATP and I 5HT characteristics ( Fig. 3D; Table 1). From protocol-5, the trace evoked by 120 mV was fit with a standard (i.e., single or double) exponential function using the following: A1 exp½Àðt À kÞ=1 C: Fitting and decay tau (t ; ms) calculation was performed using pCLAMP10.6 software (Fig. 1E). Shape of current (I), including presence or absence of "spike-like" feature at steps to 0 and 120 mV, was an important clustering variable (Figs. 3F, 4C). Clustering is based on at least two unique features for a specified cluster (Table 2). An approach for generation of clustering parameters described in detail in Results and in the previous publication (Patil et al., 2018).
Images were acquired using a Keyence BZ-X810 Allin-One Fluorescent Microscope (Keyence) or a Nikon Eclipse 90i microscope (Nikon Instruments) equipped with a C1si laser scanning confocal imaging system. Images were processed with NIS-elements (Nikon Instruments) and Adobe Photoshop CC 2019 software. Gain setting was constant during acquisition, and it was established on no primary control slides. Control IHC was performed on tissue sections processed as described but either lacking primary antibodies or lacking primary and secondary antibodies. Cell counts from IHC images acquired as Z-stuck were performed using ImageJ software. Cells positive for WGA and each marker as well as the combinations of markers were counted. We used three or four independent mice (n = 3-4) to generate sections and counted three to five sections per mouse. Mean values from these three to five sections represented n of 1.

Statistical analysis
GraphPad Prism 8.0 (GraphPad) was used for statistical analyses. Data in the figures are mean 6 SEM, with n referring to the number of mice used for IHC and the numbers of analyzed recorded cells. Differences between IHC and electrophysiologically characterized groups were assessed by unpaired t test or regular one-way ANOVA with Tukey's post hoc tests, each column was compared with all other columns. A difference is accepted as statistically significant when p , 0.05. Interaction F ratios and the associated p values are reported.

Results
Sensory neurons have previously been identified and classified using multiple methods including, back-tracing from the target tissue (da Silva Serra et al., 2016), sensory neuron reporter mice, patch-clamp recording and classification according to AP properties, sensitivity to algesic agents and appearances of a variety of voltage-gated currents (Xu et al., 2010;Li et al., 2011;Patil et al., 2018) and IHC (Patil et al., 2018). Described in Materials and Methods, back-tracing from MM will label TG neurons innervating MM fibers, tendon, muscle-tendon junction, and massteric fascia. This approach could have two drawbacks. First, along with these MM structures, diffusion of WGA injected into MM led to labeling of adjusted tissues such as masseteric nerve fibers and the subcutaneous layer of facial skin (Fig. 13A,B). Second, WGA produces nonuniform size-dependent labeling of neurons (Robertson, 1990). This nonuniform labeling of TG neurons is unlikely miss entire neuronal group since there is a variation in neuronal sizes within every TG neuronal group. Patch clamp experiments from these WGA-labeled MM TG neurons will yield data on 12 variables: cell size, RMP, AP width (dB), characteristic features of AP shapes, AHP-peak, AHP 80 , responsiveness to 5HT, ATP, capsaicin (CAP), MO, and Ment, t (tau) from fitting of voltage-gated currents (I), and shape of these I, including presence or absence of a spike-like feature on outward portions of these voltage-gated currents (Table 1).

Figure 2.
Recordings from TRPV1 1 MM TG neurons. A, In MM TG neurons belonging to the group S3, a single AP in WGA 1 /TRPV1 1 neuron is generated by current pulse (protocol-1), which has characteristic hump on down stroke phase of AP. B, In MM TG neurons belonging to the group S4, a single AP in WGA 1 /TRPV1 1 neuron has distinctive deflection on down stroke phase of AP. C, Current-evoked AP train from a WGA 1 /TRPV1 1 TG neuron belonging to the group S4. Depictured AP train is evoked by a 50-pA step lasting 1 s. D, In WGA 1 /TRPV1 1 TG neurons belonging to the group S3, currents were generated by waveforms from Figure 1E. E, In WGA 1 /TRPV1 1 TG neurons belonging to the group S4, currents generated by waveforms from Figure 1E do not have inward component.

IB4 1 MM TG sensory neuronal groups
Cultured TG neurons from wild-type male mice backtraced with WGA-488 from MM were stained with IB-4-555 for 0.5-4 h (Patil et al., 2018). WGA 1 cells considered IB-4 1 , when strong and clear plasma membrane staining was detected (Fang et al., 2006;Fig. 1A). These strongly stained IB-4 1 neurons have been considered as non-peptidergic (Stucky and Lewin, 1999). Seventy-one WGA 1 / IB-4 1 MM TG neurons (Fig. 1A) were recorded with sequential protocols as described in the Materials and Methods. All but three IB-4 1 MM TG neurons could be assigned to one of two clusters (S1 and S2; Table 1). S1 and S2 have many similar features such as size, RMP, broad AP (i.e., high dB values; Fig. 1B,C), AHP 80 , time to AHP peak, non-responsiveness to ATP, 5HT and CAP, capability to produce evoked AP train (Fig. 1D) and t values (Table 1). S2 IB-4 1 MM TG neurons are distinct from S1 because of the following three findings: only S2 neurons were responsive to MO (25 mm; TRPA1 agonist; Fig. 3A); S2 has deflection on AP (Fig. 1B), while S1 has bow (Fig. 1C); and unlike S1 but like S4, current (I) in S2 neurons had no inward component (compare Figs. 1E and 2E). S1 and S2 are different from other types of MM TG neurons as we observed staining with IB-4, but no CGRP 1 labeling (i.e., IB-4 1 /CGRPneurons). Overall, we observed two subtypes of non-peptidergic smallsized TG neurons labeled after back-tracing with WGA from MM.

TRPV1 1 MM TG sensory neuronal groups
Fifty-five TRPV1 1 MM TG neurons (WGA 1 /TRPV1 1 ) were recorded from TRPV1-GFP reporter mice (Patil et al., 2018). Two groups (S3 and S4) were delineated. S3 and S4 had several common electrophysiological properties: size, RMP, broad AP, AHP 80 , time to AHP peak ( Fig.  2A,B), non-responsiveness to ATP and 5HT, capabilities to produce evoked AP train (Fig. 2C) and t values (Table  1). We found S3 TRPV1 1 MM TG neurons were differentiated from S4 because of the following: only S3 neurons were responsive to MO (Fig. 3A); S3 had a hump (more pronounced than deflection) on AP ( Fig. 2A), while S4 had a deflection (Fig. 2B); and unlike S4, current (I) in S3 neurons had a large inward component (compare Fig. 2D,E). S3 and S4 are dissimilar from other types of TG neurons back-traced from MM in expression of TRPV1 and responsiveness to CAP (Table 2). In summary, there are two subtypes of TRPV1 1 small-sized TG neurons labeled by WGA injected into MM.

CGRP 1 MM TG sensory neuronal groups
A total of 122 CGRP 1 MM-TG neurons were recorded and analyzed. We selected approximately equal numbers of small-sized (,30 pF) and medium-to-large-sized (.30 pF) neurons for recording. A majority (48 from 62) of small-sized (,30 pF) WGA 1 /CGRP 1 neurons were CAP responsive and according to their properties, could be classified as either S3 or S4 (Table 1). This indicated that all TRPV1 1 MM TG neurons were peptidergic small-sized neurons. Fourteen small-sized WGA 1 /CGRP 1 neurons, clustered as S5, had unique characteristics: straight shape on falling phase of AP (Fig. 3B), small 5HT-evoked current (Fig. 3D), no response to CAP, MO, and Ment (TRPM8 agonist), and "box"-shaped current without an inward component (Fig. 3E). Moreover, S5 neurons readily generated an AP train even after 50-pA current injection (Table 1).
Sixty recorded and analyzed medium-to-large-sized CGRP 1 MM TG neurons could be divided into two groups, M1 and M2. Unlike all small-to-large-sized (i.e., S1-S5) and other medium-to-large-sized (i.e., M3 and M4) MM TG neurons, M1 and M2 possessed a large 5HTevoked inward current ( Fig. 3D; Table 1). We have additionally recorded from 24 WGA 1 /5HT3a 1 neurons cultured from 5HT3a-GFP reporter mouse TG. All recorded neurons were classified as either S5, M1 or M2, and responded to 5HT (Fig. 3D). Unlike small-sized MM TG neurons (S1-S5 groups), large current injections were required to produce AP train, which was detected only in 10-20% of M1 and M2 neurons ( Table 1). The I (current) recorded from M1 was like S5 neurons, having the lowest t values among all MM TG neuronal groups (one-way ANOVA; F (8,261) = 5.631; p , 0.0001; Fig. 3E; Table 1). The most distinct features of M2 compare to M1 neurons was their significantly narrower AP (t test; t = 5.149 df = 109; p , 0.0001; Fig. 3C) and substantially different shape of current (I) with spike and large inward component (compare Fig. 3E,F). These characteristics indicate that M1 and M2 belong to medium-sized peptidergic TG MM neurons.
Next, using reporter mice, we recorded and analyzed 65 trkC 1 , 32 PV 1 , and 35 medium-to-large sized Nav1.8 1 MM TG neurons. This information revealed that trkC was expressed by S5, M2-M4 neurons; PV was present in M2-M4 groups; and Nav1.8 was mainly detected in S1- A, Representative AP from a MM TG trkC 1 neuron belonging to the M3 group. B, Representative AP from a MM TG trkC 1 neuron belonging to the M4 group. C, Typical I produced from the M3 group neurons recorded either from a MM TG trkC 1 or Nav1.8neuron. Characteristic spike is indicated by green arrow. D, Typical I produced from the M3 group neurons recorded from a MM TG trkC 1 or Nav1.8neuron. Names of neuronal groups are specified above traces on the A-D panels. The magnitude (vertical) and time (horizontal) scale bars are presented for the A-D panels. S4 groups. Medium-to-large sized Nav1.8 1 MM TG neurons had weaker YFP signal than Nav1.8 1 small-sized neurons. A bulk majority of medium-to-large sized Nav1.8 1 MM TG neurons belonged to the M1 group, but four of 35 were classified into the M2 group. Since M1 and M2 neurons were recorded among Nav1.8 1 and Nav1.8neurons, this suggests that they had weak Nav1.8 expression, which was sufficient to drive Ai32 reporter in some, but not for all M1 and M2 neurons. Figure 13C summarizes clustering of TG neurons labeled by back-tracing from MM. These clusters were generated on this basis of electrophysiological properties obtained from recordings on TG neurons isolated from different reporter mouse lines. Unlike previously characterized L3-L5 DRG neurons, which innervate the skin and muscle of legs, TG neurons labeled by backtracing from MM did not have small-sized TRPV1 1 / CGRP -(somatostatin 1 ) neurons (Usoskin et al., 2015;Sharma et al., 2020), TRPM8 1 (cold thermoceptors) neurons (Sharma et al., 2020) and strongly responding to ATP MrgprA3 1 neurons (Usoskin et al., 2015;Patil et al., 2018). MM TG neurons also did not have multiple small-sized peptidergic groups (Patil et al., 2018;Zeisel et al., 2018). Expectedly, MM TG lacked neuronal groups associated with innervation of hairs: C-LTMR and Ad -LTMR, which are Nav1.8 1 /CGRP -/trkC - (Usoskin et al., 2015) and have distinct AP (Patil et al., 2018). Finally, MM TG neurons do not have proprioceptors (Sharma et al., 2020), which are located in a brain stem region of the TG system. Overall, MM TG neuronal groups were found to be substantially different from the well-characterized L3-L5 DRG neuronal clusters, and S5 and M2 groups have no analogs among L3-L5 DRG neurons (see Discussion).

TG sensory afferent nerve types innervating masseteric muscle fibers and tendons
The masseteric nerve enters the deep portion of MM and is divided into many branches within MM (Fig. 13A,B). These branches eventually supply the sensory afferent nerves that innervate MM fibers, tendon, muscle-tendon junction, and masseteric fascia. Injection into MM could lead to contamination and/or diffusion of WGA into subcutaneous tissue (Fig. 13B). Additionally, such an injection could label sensory afferent fibers in the main trunk of masseteric nerve (Fig. 13A,B). Using IHC, we examined the distribution of different sensory fiber types within MM, as well as innervation of muscle fibers, tendon, muscletendon junctions, and fascia surrounding whole MM.
The main trunk of the masseteric nerve brunches out from the mandibular nerve and enter the MM in the sigmoid fascia area (Kim et al., 2010; Fig. 13A,B). We found this area of masseteric nerve trunk contained CGRP 1 / NFH -(yellow arrows; S3-S5 groups), CGRP 1 /NFH 1 (M1 and M2 groups; Fig. 9A,A', cyan arrows) and CGRP -/ NFH 1 (M3 and M4 groups; Fig. 9A,A', white arrows) sensory nerves. The masseteric nerve trunk then descends into MM between the middle and deep parts of muscle fibers and then divides into many nerve branches in the posterosuperior, posteroinferior, anterosuperior, and anteroinferior directions (Luo et al., 1991;Kim et al., 2010;Fig. 13A,B). We observed that multiple branches of CGRP 1 and NFH 1 nerve bundles reached the deep and middle layers of the muscle and junctions of muscle fibers and tendon (Fig. 9B). Distribution of CGRP 1 and NFH 1 nerve bundles and individual fibers occurred throughout MM. However, they mainly concentrate in tendon and at junctions between superficial, middle, and deep parts of MM (Fig. 9B,C,C'). Additionally, these fibers were found to innervate masseteric fascia, which surrounds the whole MM (Fig. 9D).
To further investigate this unexpected finding, we examined MM distribution of nerves with peptidergic CGRP 1 / NFHmarkers such as trpV1 1 , Nav1.8 1 , and pgp9.5 1 ; and NFHmarkers such as trkC 1 and PV 1 . Moreover, to evaluate whether these findings are translated across species, MM fiber types were labeled in the marmoset tissue. Pgp9.5 is marker for all types of sensory nerves, and could highlight location of non-peptidergic unmyelinated nerve fibers (Tokushige et al., 2006). Analysis 8-16 slides with MM sections from male mice showed that pgp9.5 1 nerves innervating muscle fibers (Fig. 11A,A'), tendon of MM (Fig. 11B), and masseteric fascia (Fig. 11C) are always co-labeled by NFH (cyan arrows on all panels). Muscle fibers of male marmosets were predominantly innervated by pgp9.5 1 /NFH 1 nerves (Fig. 11D,D', cyan arrows). However, pgp9.5 1 /NFHnerves (Fig. 11D,D', white arrows), which likely represent C-fibers, were seldomly detected on one of two MM sections observed with 10Â objective. These findings imply that male mouse MM fibers, tendon and fascia are almost solely innervated by A-fiber myelinated fibers, while C-fiber unmyelinated fibers are rarely encountered in MM fibers of marmosets.
TRPV1 marks S3 and S4 groups of MM TG neurons containing peptidergic unmyelinated fibers. TRPV1 1 nerves were not detected within masseteric muscle fibers (Fig. 11A,A'), tendon or fascia of MM (data not shown). However, some nerve branches of masseteric nerve located along muscle fibers contain TRPV1 1 sensory fibers, which was also labeled by pgp9.5 1 (Fig. 11B,B', cyan arrows). Nav1.8 1 fibers were highlighted by isolation of MM from Nav1.8-cre/Ai32-YFP mice. Electrophysiology recording showed that strong Nav1.8-YFP signal is detectable only in small neurons (S1-S4 groups), while M1 has weak Nav1.8 presence and S5 and M2-M4 have none. Hence, Nav1.8 1 sensory fibers were not detected within muscle fibers and tendon. However, certain masseteric nerve branches contained Nav1.8 1 fibers (Fig. 11C,C', red arrow). Interestingly, we have detected multiple nerve branches adjusted or inside MM expressing both pgp9.5 and NFH, but not showing Nav1.8 (Fig. 11C,C', yellow arrows). Expression patterns of CGRP 1 , pgp9.5 1 , Nav1.8 1 , and NFH 1 nerves within MM indicate that muscle fibers, tendon and fascia are innervated by M1-M4 and maybe S5 TG MM neurons. In concordance with this, trkC, which is a marker for S5 and M2-M4 MM TG neurons (Fig. 13C), was detected in many NFH 1 fibers (Fig. 11D,D', cyan arrows). However, unlike trkC, anti-PV antibodies produced only weak labeling in masseteric nerve branches (data not shown). Overall, our data suggest that muscle fibers, tendon, and fascia of MM in mice and marmosets are almost exclusively innervated by S5 and M1-M4 nerves, which are mainly distributed in tendon and at junctions of superficial, middle, and deep parts of MM. Unmyelinated IB-4 1 , Nav1.8 1 , and TRPV1 1 nerves (S1-S4 neurons) are in a subset of masseteric nerve branches, but do not terminate in muscle fibers, tendon, and fascia of mice MM (Fig.  13A,B). Additionally, it could not be excluded that WGA injected in MM could defuse into adjusted structures such as skin and especially, subcutaneous tissues, which contain unmyelinated C-fibers (see Discussion). Figure 9. Location of CGRP 1 and NFH 1 TG sensory neuronal afferent fibers in MM. A, A', Expression of CGRP (green) and NFH (blue) fibers in a main trunk of the masseteric nerve located near MM. Objective is 40Â. CGRP 1 /NFHnerves are marked with yellow, CGRP 1 /NFH 1 with cyan, and CGRP -/NFH 1 with white arrows. Red arrow shows MM fibers. B, Expression of CGRP (red) and NFH (green) nerves in muscle fibers and MM tendon. Objective is 10Â. Red arrow shows muscle fibers and blue points on tendons. Nerve bundles are marked with yellow arrows and individual nerves with cyan arrows. C, Expression of CGRP (red) and NFH (green) nerves in MM. Objective is 2Â. A red arrow shows muscle fibers. C', High magnification (objective is 20Â) of the image within a white rectangle on the panel C. On the panels C, C', nerve bundle is marked with a yellow arrow and individual nerves with cyan arrows. D, Expression of Nav1.8-Ai32-YFP (green), CGRP (red), and NFH (blue) nerves in MM fascia area, which is indicated by white double arrowed line. Objective is 10Â. White horizontal bar shows 20-or 100-mm scale for each panel.

Discussion
Mechanisms underlying masticatory myofascial pain are largely unknown. Masticatory myofascial pain involves MM in several conditions, including TMD and after some restorative surgeries in the head and neck. In this respect, understanding jaw muscle and particularly MM innervation is critical for dissecting cell-specific mechanisms controlling development of chronic masticatory myofascial pain. This understanding will eventually allow for the effective management of TMD pain (Dinsdale et al., 2021). In addition, masseteric nerve is affected during certain restorative surgeries and during transplantations/transfers, which are the most often performed cranial nerve manipulation technique used to treat patients with facial paralysis (Corcoran and Goldman, 2021). In this study, using backtracking from MM with WGA, reporter mice driven by promoters of well-characterized sensory neuronal markers, patch clamp electrophysiology and IHC, we identified and characterized types of TG sensory neurons innervating MM. This approach was successfully employed to identify and characterize neuronal groups in L3-L5 DRG and TG (Petruska et al., 2000;Xu et al., 2010;Patil et al., 2018). We clustered most recorded MM TG neurons into nine clusters ( Fig. 13C; Table 2). One to two percent of recorded neurons did not fit these nine clusters. For example, MM TG neurons did not respond to ATP (Table 1), but from .350 analyzed neurons, seven responded to ATP.
There is not clear consensus on how the sensory neurons between DRG and TG differ. There is also not enough data on the biochemical differences between sensory neurons innervating the MM. Nevertheless, there is agreement that somatosensory neurons in different sensory ganglia have distinct anatomic, functional, physiological, and pathophysiological characteristics (Li, 2007;Belmonte and Viana, 2008;Gambeta et al., 2020). Thus, DRG are located in the intervertebral foramina at spinal levels that mainly innervate tissues within the trunk, legs and feet, while TG rest in Meckel's cave and innervate the head and neck area (Vermeiren et al., 2020). Another anatomic distinction is proprioceptors with Aa-fibers are located in DRG but are outside TG in the mesencephalic TG nucleus (Jerge, 1963;Dubner, 1986). Since physiological function of tissues in the trunk, legs and feet are quite distinct from tissues in head and neck area, it is presumed that DRG and TG sensory neurons have distinct, specialized, and unique sets of proteins/mRNA. With the advent of RNA-seq, data from sorted sensory neurons and single-cells described some of these differences in recent years (Usoskin et al., 2015;Lopes et al., 2017;Nguyen et al., 2019;Zheng et al., 2019;Mecklenburg et al., 2020;Renthal et al., 2020). However, these sequencing data indicate that the differences between sensory neurons in DRG and TG are not dramatic and lack clear functional distinctions (Lopes et al., 2017;Mecklenburg et al., 2020;Sharma et al., 2020). These differences between DRG and TG neurons could be because of analysis of sensory neuronal profiles in whole ganglia, but not in specific subsets of sensory neurons innervating defined tissues (such as muscle, skin, tooth, dura, etc.). Considering that neuronal subsets may have distinct origins and may therefore be guided to different targets, innervating different tissues (Hockley et al., 2019), we focused our study on a specific subset of TG neurons innervating MM.
Ideally, our data need to be compared with DRG neurons innervating hind/fore limb muscle. However, this information is not available. Single-cell data generated by several independent studies imply that difference between neuronal groups innervating limb skin and muscle maybe not substantial (Usoskin et al., 2015;Sharma et al., 2020). Thus, our data indicate that 20-25% of L3-L5 DRG neurons were labeled by WGA injected into limb muscles (Mecklenburg et al., 2020). Single-cell sequencing of thousands of sensory neurons clustered major phenotypic differences of DRG groups innervating the skin and muscle, for example, proprioceptors were clustered as a separate group (Usoskin et al., 2015;Sharma et al., 2020). Taken all these points into consideration, we compared our results with published data from whole L3-L5 DRG. S1 and S2 groups, which are Nav1.8 1 , but not CGRP 1 , are classic CAP-unresponsive IB-4 1 non-peptidergic sensory neurons (Stucky and Lewin, 1999). Unlike L3-L5 DRG neurons , MM TG IB-4 1 neurons were not MrgprD 1 . Functional studies on DRG neurons distinguished two types of skin innervating IB-4 1 / MrgprD 1 neurons: C-polymodal nociceptors (PMN) and C-mechano-nociceptors (C-MN; Liu et al., 2012;. For MM TG neurons, it is not clear what functions could be assigned to S1 and TRPA1 containing S2 groups. S3 and S4 groups containing CGRP, TRPV1 and Nav1.8 are likely C-PMN-like with similarities to DRG neurons (Usoskin et al., 2015;Patil et al., 2018;Sharma et al., 2020). The last small-sized MM TG neuronal group characterized was S5. It had unique electrophysiological properties (Table 1) and we observed expression of CGRP and trkC, but not Nav1.8 in S5 group MM TG neurons. There is no analog for S5 neurons in L3-L5 DRG; and it is unknown whether S5 express NFH and belong to nociceptors or LTMR. Unlike DRG neurons innervating skin, MrgprA3 1 (C-PMN; Han et al., 2013), CGRP -/ TRPV1 1 (also known as somatostatin 1 ; Usoskin et al., 2015), and TH 1 (also known as C-LTMR; Seal et al., 2009) groups were not found among MM TG neurons.
As it was previously reported (Sato et al., 2018), medium-sized neurons are most abundant among MM TG neurons (Fig. 13C). Moreover, in L3-L5 DRG innervating either skin or muscle, all CGRP 1 neurons were found inside of the Nav1.8 1 subset (Patil et al., 2018), while S5, M1, and M2 MM TG neuronal groups expressed CGRP but little-to-no Nav1.8 (Fig. 13C). M1 and M2 mediumsized MM TG peptidergic neurons produced large inward currents on application of 5HT. This property distinguished M1 and M2 from other MM TG neurons. M1 Figure 11. MM TG neuron types innervating MM in mice and common marmosets. A, A', Expression of pgp9.5 (red) and NFH (green) in male mouse masseteric muscle fibers. Objective is 10Â. Pgp9.5 1 /NFH 1 nerves are shown with cyan arrows. B, Expression of pgp9.5 (red) and NFH (blue) in male mouse MM tendon. Objective is 10Â. Muscle fibers are shown by a red arrow, and tendons by a green arrow. Pgp9.5 1 /NFH 1 nerves are shown with cyan arrows. C, Expression of pgp9.5 (red) and NFH (blue) in male mouse masseteric fascia. Objective is 20Â. Muscle fibers are shown by a red arrow, and fascia layer by a double-headed white arrow. Pgp9.5 1 /NFH 1 nerves are shown with cyan arrows, and a Pgp9.5 -/NFH 1 nerve with a yellow arrow. D, D', Expression of pgp9.5 (green) and NFH (red) in 11-year-old male marmoset masseteric muscle fibers. Objective is 10Â. Pgp9.5 1 /NFH 1 nerves are shown with cyan arrows, and Pgp9.5 1 /NFHnerves with white arrows. White horizontal bar shows 20-mm scale for each panel.
Distribution of nerves in masseteric muscles has been studied in rats, rabbits, pig and cats (Gottlieb et al., 1984;Herring et al., 1989;Luo et al., 1991;Peker et al., 2001). It was shown that the masseteric nerve is subdivided into at least three to five subbranches within MM (Fig. 13A,B). Myelin sheets surrounding the unmyelinated nerve fibers disappeared as the fibers terminate (Herring et al., 1989).
Our data shows that unmyelinated and myelinated fibers travel together in a main trunk of the masseteric nerve (Fig. 9A,A'). Similarly, nerve branches of the masseteric nerve contain either both unmyelinated and myelinated fibers or only myelinated fibers (Fig. 12C,C'). It was reported that distribution of different types of fibers occur throughout MM and are independent of fiber type (Herring et al., 1989). First, the data indicate that afferent fibers are concentrated in tendons and at junctions between deep, middle, and superficial parts of MM (Fig. 9B,C,C'). Second, importantly, male mouse muscle fibers, tendons, and masseteric fascia of MM were almost exclusively innervated by myelinated NFH 1 afferent nerves (Figs. 9B,C, C',D, 10, 11A-C). Likewise, male marmoset MM were predominantly innervated by myelinated NFH 1 afferent nerves (Fig. 9D,D'). Mouse MM TG neurons with myelinated fibers belong to M1-M4 groups, and maybe group S5 as well. IHC could detect unmyelinated MM TG neurons, which could be identified as Nav1.8 1 /NFH -, TRPV1 1 , and/or pgp9.5 1 /NFHafferent nerves, only within a subset of masseteric nerve branches (Fig. 12A-D'). It is not clear where these unmyelinated fibers terminate, since they were not detected by IHC in MM fibers, MM tendons and masseteric fascia. Accordingly, WGA injection into MM could label small TG neurons (S1-S4) found in the main trunk or sub-branches of the masseteric Figure 13. Schematic abstract of results. A, Schematic of different parts of masseteric muscle (labeled) and branches of masseteric nerve (labeled). B, Schematic of coronal section of MM with adjusted tissues. These adjacent tissues and MM components are labeled. On the panels A, B, blue lines and dots mark A-fibers, and red lines and dots represent C-fibers in mandibular and masseteric nerves and in their branches. C, Schematic TG neuronal groups (S1-S5, M1-M4) innervating MM. Expressions of neuronal markers are specified. Percentage of WGA 1 for each neuronal group based on IHC is shown. Putative functions of MM TG neuronal groups are suggested. A-HTMR are medium-sized A-nociceptors or high threshold mechanoreceptors. A-LTMR are medium-tolarge-sized A-low threshold mechanoreceptors. Putative functions of S5 and M2 are ambiguous, do not map onto previously described classes of DRG neurons, and marked as "unknown." nerve, and/or adjacent to MM subcutaneous tissues and skin via WGA diffusion (Fig. 13A,B).
In conclusion, our data show that MM is almost exclusively innervated by myelinated nerves that can be divided into S5, M1-M4 neuronal groups (Fig. 13A-C). WGA injections into MM also revealed small sized TG neurons (S1-S4), nerve fibers from which are within the masseteric nerve or terminate in adjusted to MM structures (Fig. 13B,  C). Properties of these MM TG neuronal groups have similarities and many distinctions from the well-described and characterized L3-L5 DRG neurons innervating limb skin and muscles (Petruska et al., 2000;Usoskin et al., 2015;Patil et al., 2018;Sharma et al., 2020). These differences could imply distinct functional consequences during naive and pathologic conditions affecting MM. It also suggests that TG MM neurons may contain a discrete subset of receptors and channels, which could uniquely regulate sensitization of MM TG neurons during masticatory myofascial pain. Finally, our novel report on the innervation of MM in a non-human primate species, the common marmoset, is important for many reasons. Because of the close relatedness of nonhuman primates and humans, delineation of the MM in the marmoset is important to support its use as a potential animal model MM-related pathology. Further, identification of similarities between marmoset and mouse MM further supports the use of both species as preclinical models for development of intervention approaches in this area.